Implementing stochastic networks using magnetic tunnel junctions

Abstract

Embodiments of the invention relate to implementing a probabilistic graphical model (PGM) using magnetic tunnel junctions (MTJs). One embodiment comprises a memory array of magnetic tunnel junctions and a driver unit for programming the memory array to represent a probabilistic graphical model. The magnetic tunnel junctions are organized into multiple subsets of magnetic tunnel junctions. The driver unit selectively applies an electrical pulse to a subset of magnetic tunnel junctions to program information representing a probabilistic belief state in said subset of magnetic tunnel junctions.

What is claimed is: 
     
       1. A system comprising:
 at least one processor; and 
 a non-transitory processor-readable memory device storing instructions that when executed by the at least one processor causes the at least one processor to perform operations including:
 maintaining network definition data relating to a probabilistic graphical model; and 
 programming a memory array of magnetic tunnel junctions based on the network definition data, wherein the programming comprises:
 encoding a first subset of magnetic tunnel junctions of the memory array with a known probability value representing an observed probabilistic belief state of a first node of the probabilistic graphical model by determining a first magnitude and a first pulse width for a first electrical pulse based on the network definition data, and applying the first electrical pulse with the first magnitude and the first pulse width to the first subset of magnetic tunnel junctions; and 
 encoding a second subset of magnetic tunnel junctions of the memory array with an estimated probability value representing an estimated probabilistic belief state of a second node of the probabilistic graphical model by determining a second magnitude and a second pulse width for a second electrical pulse based on an inference algorithm and the network definition data, and applying the second electrical pulse with the second magnitude and the second pulse width to the second subset of magnetic tunnel junctions. 
 
 
 
     
     
       2. The system of  claim 1 , wherein the estimated probabilistic belief state of the second node of the probabilistic graphical model is estimated based in part on the observed probabilistic belief state of the first node of the probabilistic graphical model. 
     
     
       3. The system of  claim 1 , wherein the probabilistic graphical model comprises multiple nodes interconnected via multiple edges. 
     
     
       4. The system of  claim 3 , wherein the network definition data comprises edge connectivity information for the multiple nodes. 
     
     
       5. The system of  claim 3 , wherein each node of the multiple nodes has either an observed probabilistic belief state or an estimated probabilistic belief state. 
     
     
       6. The system of  claim 5 , wherein the network definition data comprises evidence information for each node of the multiple nodes with an observed probabilistic belief state. 
     
     
       7. The system of  claim 6 , wherein the evidence information comprises, for each node of the multiple nodes with an observed probabilistic belief state, a known probability value representing the observed probabilistic belief state. 
     
     
       8. The system of  claim 1 , wherein determining a second magnitude and a second pulse width for a second electrical pulse based on an inference algorithm and the network definition data comprises:
 determining the estimated probability value representing the estimated probabilistic belief state of the second node of the probabilistic graphical model based on the inference algorithm and the network definition data. 
 
     
     
       9. The system of  claim 3 , wherein the multiple edges are arranged in a directed acyclic graph. 
     
     
       10. A method comprising:
 maintaining network definition data relating to a probabilistic graphical model; and 
 programming a memory array of magnetic tunnel junctions based on the network definition data, wherein the programming comprises:
 encoding a first subset of magnetic tunnel junctions of the memory array with a known probability value representing an observed probabilistic belief state of a first node of the probabilistic graphical model by determining a first magnitude and a first pulse width for a first electrical pulse based on the network definition data, and applying the first electrical pulse with the first magnitude and the first pulse width to the first subset of magnetic tunnel junctions; and 
 encoding a second subset of magnetic tunnel junctions of the memory array with an estimated probability value representing an estimated probabilistic belief state of a second node of the probabilistic graphical model by determining a second magnitude and a second pulse width for a second electrical pulse based on an inference algorithm and the network definition data, and applying the second electrical pulse with the second magnitude and the second pulse width to the second subset of magnetic tunnel junctions. 
 
 
     
     
       11. The method of  claim 10 , wherein the estimated probabilistic belief state of the second node of the probabilistic graphical model is estimated based in part on the observed probabilistic belief state of the first node of the probabilistic graphical model. 
     
     
       12. The method of  claim 10 , wherein the probabilistic graphical model comprises multiple nodes interconnected via multiple edges. 
     
     
       13. The method of  claim 12 , wherein the network definition data comprises edge connectivity information for the multiple nodes. 
     
     
       14. The method of  claim 12 , wherein each node of the multiple nodes has either an observed probabilistic belief state or an estimated probabilistic belief state. 
     
     
       15. The method of  claim 14 , wherein the network definition data comprises evidence information for each node of the multiple nodes with an observed probabilistic belief state. 
     
     
       16. The method of  claim 15 , wherein the evidence information comprises, for each node of the multiple nodes with an observed probabilistic belief state, a known probability value representing the observed probabilistic belief state. 
     
     
       17. The method of  claim 10 , wherein determining a second magnitude and a second pulse width for a second electrical pulse based on an inference algorithm and the network definition data comprises:
 determining the estimated probability value representing the estimated probabilistic belief state of the second node of the probabilistic graphical model based on the inference algorithm and the network definition data. 
 
     
     
       18. The method of  claim 12 , wherein the multiple edges are arranged in a directed acyclic graph. 
     
     
       19. A non-transitory computer readable storage medium including instructions to perform a method comprising:
 maintaining network definition data relating to a probabilistic graphical model; and 
 programming a memory array of magnetic tunnel junctions based on the network definition data, wherein the programming comprises:
 encoding a first subset of magnetic tunnel junctions of the memory array with a known probability value representing an observed probabilistic belief state of a first node of the probabilistic graphical model by determining a first magnitude and a first pulse width for a first electrical pulse based on the network definition data, and applying the first electrical pulse with the first magnitude and the first pulse width to the first subset of magnetic tunnel junctions; and 
 encoding a second subset of magnetic tunnel junctions of the memory array with an estimated probability value representing an estimated probabilistic belief state of a second node of the probabilistic graphical model by determining a second magnitude and a second pulse width for a second electrical pulse based on an inference algorithm and the network definition data, and applying the second electrical pulse with the second magnitude and the second pulse width to the second subset of magnetic tunnel junctions. 
 
 
     
     
       20. The computer readable storage medium of  claim 19 , wherein the estimated probabilistic belief state of the second node of the probabilistic graphical model is estimated based in part on the observed probabilistic belief state of the first node of the probabilistic graphical model.